Innovation and firm value: An investigation of the changing role of patents, 1985-2007

(1)

ResearchPolicy42 (2013) 1496–1510

ContentslistsavailableatSciVerseScienceDirect

Research

Policy

jou rn a l h o m e pa g e :w w w . e l s e v i e r . c o m / l o c a t e / r e s p o l

Innovation

and

ﬁrm

value:

An

investigation

of

the

changing

role

of

patents,

1985–2007

Sharon

Belenzon

a,1

_,

_Andrea

_Patacconi

b,∗

a_Fuqua_School_of_Business,_Duke_University,_USA

b_University_of_Aberdeen_Business_School,_UK

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received17March2010

Receivedinrevisedform29April2013 Accepted3May2013

Available online 16 June 2013

JELclassiﬁcation: O31 O32 O16 Keywords: Firmvalue

Patentportfoliovaluation Patentpolicy

a

b

s

t

r

a

c

t

Thispaperexamineshowtherelationshipbetweenfirmvalueandpatent-basedindicatorsofinventive activityhaschangedovertime.Weusedatafrommorethan33,000mergersandacquisitionsdeals between1985and2007,anddistinguishbetweenAmerican(USPTO)andEuropean(EPO)patents.Our resultsindicatethatovertimeEPOpatentshavebecomethedominantindicatorofinnovativeactivity, whileUSPTOpatentshavenoeffectonfirmvalueneartheendofthesampleperiod.Theresultsarerobust tocontrollingforcitationsandareespeciallystrongforsmallfirms,forfirmsoperatinginthedrugand chemicalindustries,andwhentargetandacquiringfirmsoperateindifferentindustriesorcountries.

1. Introduction

In evaluating potentialtargets for mergers and acquisitions

(M&A),acquiringﬁrmsmusttakeseveralfactorsintoaccount:the

valueofthetargetcompany’stangibleandintangibleassets,the

potentialforsynergiesbetweenthemergingﬁrms,cost-of-capital

andtaxconsiderations,andmore.Afactorthathasbecome

increas-inglyimportantin recent years isintellectual property.A large

patentportfoliocanpositivelyinﬂuenceaﬁrm’svaluationinM&A

foranumber ofreasons.Patentsarelegaltitlesthatgrant their

ownersatemporarymonopolypowerovertheuseofaninvention.

Thus,theycanbeusedtoenforceexclusivity,toextract

licens-ingincome,orasabargainingchipinnegotiations.Wereferto

theseusesofintellectualpropertyasthemonopolyorlegalvalueof

patents.Patentscanalsobecorrelatedwithﬁrmvaluebecausethey

provideinformation aboutthequalityof inventionsand overall

“corporatetechnologicalstrength”(Long,2002;Narinetal.,1987).

Patentingisexpensive,thusonlyinventionsthataredeemed

suf-ﬁcientlyvaluablearepatented.Moreover,examinersmustcheck

∗_{Corresponding}_author_at: _University_of_Aberdeen_Business_School,_Edward

WrightBuilding,DunbarStreet,AberdeenAB243QY,UK.Tel.:+4401224273423.

E-mailaddresses:sharon.belenzon@duke.edu(S.Belenzon),

a.patacconi@abdn.ac.uk(A.Patacconi).

1 _Address:_Duke_University:_The_Fuqua_School_of_Business,₁₀₀_Fuqua_Drive, Durham,NC27708,USA.Tel.:+19196607845.

thatinventionssatisfyminimumstandardsof noveltyand

non-obviousness.More broadly,largepatent portfolios maysuggest

thatﬁrmsarewell-managedandpossessvaluableR&Dcapabilities.

Thus,largepatentportfolioscansignaltechnologicalstrength.

However,recentpatentandadministrativereformshavemost

likely affected both the monopoly and the informational (or

“quality-signaling”) value of patents. There is a broad

consen-susamongacademicsandpractitionersthatpatentprotectionin

mostOECDcountrieshasbeenconsiderablystrengthenedsincethe

1980s(JaffeandLerner,2004;Lerner,2002).Importantstepsinthis

directionhavebeenthecreationin1982ofa“pro-patent”Courtof

AppealsintheU.S.,thesubsequentupwardharmonizationof

intel-lectualpropertyprotectionstandardsviainternationalagreements,

andtheextensionofpatentabilitytonewsubjectmattersuchas

biotechnologyandsoftware(intheU.S.,alsobusinessmethods).2

Thus,themonopolyvalueofpatentshasmostlikelyincreasedin

thelastthreedecades.

At the same time, patent examination standards appear to

havedeclined.JaffeandLerner(2004)provideseveralexamples

ofobviousornon-novelideasthatweregrantedpatentprotection,

includinga“MethodtoSwingonaSwing”,the“SealedCrustless

Sandwich”,Rambus’sdivisionalapplicationsondevelopmentsof

2_Recent_judicial_decisions,_however,_have_arguably_restricted_the_range_of busi-nessmethodsthatarepatentableintheU.S.

(2)

itsDRAMtechnology,and,perhapsmorecontroversially,one-click

businessmodels. Importantly,thedecline in examination

stan-dardsappearstohavebeenpronouncedespeciallyintheU.S.,due

toanumberofinstitutional,legalandculturalfactors(Guellecand

vanPottelsberghedelaPotterie,2007;JaffeandLerner,2004).Thus,

patent“quality”(broadly intendedtoencompassboth the

tech-nologicalandvaluedimensionsofpatentedinventions)mayhave

declinedmoreintheU.S.thaninothermajorjurisdictions.

Motivatedbytheseconsiderations,thepresentpaperexamines

howtherelationshipbetweenﬁrmvalueandpatent-based

indi-catorsofinventiveactivityhaschangedinthelastthreedecades.

We use datafrom more than 33,000 mergers and acquisitions

dealsbetween1985and2007,anddistinguishbetweenAmerican

(USPTO)andEuropean(EPO)patents.Weﬁndthatwhiletheeffect

ofEPOpatentsonacquisitionvaluehasremainedessentiallystable

overthistimeperiod,theeffectofUSPTOpatentshasdeclined

sub-stantiallyandisessentiallyzeroneartheendofthesampleperiod.3

Giventhatpatentprotectionrightswerestrengthenedduringthis

timeperiod,ourresultssuggestthatanincreasingnumberofUSPTO

patentsweregrantedforinventionsthathadlittleornoeconomic

ortechnologicalvalue.USPTOpatentsseemtobelosingtheirrole

assignalsofﬁrms’inventiveeffortsandvalue.

Therobustnessofourﬁndingsisprobedinseveralways.

Qual-itativelysimilarresultsare obtainedwhen patentsareweighed

bycitations, although citation-weighinggenerally increasesthe

importance of USPTO patents. This is expected since citations

proxyforpatentqualityandpatentqualitymaytosomeextent

beobservedby theacquiringﬁrm (Hallet al., 2005).Our

ﬁnd-ingsaredrivenbypatentsownedbysmallﬁrms.Thisisalsonot

surprisinggiventhatintellectualpropertyandsignaling

consid-erationsarearguablymostimportantinthestart-upphaseof a

ﬁrm’slife.Thereisalsosubstantialheterogeneityacrossindustries.

ThedeclineintheimportanceofUSPTOpatentsismostevidentin

drugsandchemicals,whilenosuchdeclineisfoundincomputers

andelectronics.4_We_ﬁnd_similar_results_in_{biotechnology}_and_more

traditionalsegmentsofthedrugandchemicalindustries.Thisdoes

notsupportthepresumptionthatU.S.patentqualitydeclinedmost

sharplyinnewtechnologyﬁelds(Hall,2007;Merrilletal.,2004).

Finally,EPOpatentsappeartohavebecomemoreimportantover

time,relativetoUSPTOpatents,whenacquiringandtargetﬁrms

belongtodifferentindustriesoroperateindifferentcountries.We

conjecturethat,inthesedeals,becauseacquiringﬁrmsmayﬁndit

moredifﬁculttoevaluateatargetﬁrm’stechnology,EPOpatents

maybeperceivedas“safer.”TheEPOsystemmaybeprovidingsome

qualitycertiﬁcation–acertiﬁcationwhichbecomesmore

impor-tantasuncertaintyandinformationalasymmetriesgrow(Akerlof,

1970).

Theremainderofthepaperisorganizedasfollows.Thenext

sectiondiscusses somefeatures ofthe Americanand European

patentsystemsthatmotivateourempiricalanalysis.Sections3and

4describethedataandeconometricspeciﬁcation.Section5reports

theresults.Section6concludes.

2. Backgroundandpredictions

2.1. Thechanginginstitutionalenvironment

Thereiswidespreadconsensusamongacademicsand

practi-tioners that, starting from the early1980s, various patent and

3 _However,_we_also_ﬁnd_a_large_increase_in_USPTO_patent_valuations_in_the_late 1990s.Thisincreasewasmostlikelyaconsequenceoftheexuberancethat charac-terizedofthedot-comyears.

4 _In_software,_a_technology_area_where_patenting_has_often_been_regarded_as dubi-ous,neitherUSPTOnorEPOpatentsmatterforvalueinanyofthesampleperiods.

administrativereformshave ledtoa strengtheningof thelegal

valueofpatents.IntheU.S.,akeystepinthisdirectionwasthe

cre-ationin1982ofaspecializedCourtofAppealsoftheFederalCircuit

(CAFC).AlthoughthestatedobjectiveofCongresswastoprovide

uniformityinpatentlitigationcases,thecourtsoondemonstrated

a“pro-patent”disposition(MazzoleniandNelson,1998;Merges,

1992).Speciﬁcally,thecourtwasfoundtoupholdinappealalarger

proportionofdecisionsfavorabletothepatentholder,toreverse

alargerproportionofdecisionsunfavorabletopatentholder,and

tosubstantiallyincreasetherateofpreliminaryinjunctions,

com-paredtotheprevioussystem(Gallini,2002;LanjouwandLerner,

2001).Thedeﬁnitionofpatentablesubjectmatterwasalso

broad-enedtoincludeemerging newﬁeldssuchasbiotechnologyand

software,andpatentdurationwaslengthened.Severalcountries

followedtheU.S.example,whichledovertimetoanupward

har-monizationofintellectualpropertyprotectionstandards(Guellec

andvanPottelsberghedelaPotterie,2007;Lerner,2002).

The strengthening of the legal value of patents gave ﬁrms

powerfulincentivestoseekpatentprotection(HallandZiedonis,

2001;KortumandLerner,1999).Patentapplicationsnearlytripled

between1980and2000intheU.S.,withEuropeand Japan

fol-lowingsuit.Butaspatentsbecamemorepowerfullegalweapons,

concernsabouttheeasewithwhichtheyweregrantedbeganto

grow.Observersnotedthatwaytoooftenpatentsweregranted

forideasthatwereneithernovelnornon-obvious.Some–most

notably Jaffe and Lerner (2004)– went asfar asto argue that

lowerstandardscouldendangerinnovationandprogressbecause,

ifapplicantscouldgetpatentsforobviousorexistingideas,then

patentscouldusedtoharass,seekcompensationfrom,orevenshut

downlegitimatebusinesses.Overwhelmingly,however,criticism

aboutdecliningpatentabilitystandardshasbeendirectedtoward

theAmericansystem.Severalargumentsunderpinthebeliefthat

examinationstandardsarelower(and,untilquiterecentlyatleast,

havebeengettingworse)intheU.S.thaninEuropeorJapan.

First,theUSPTOisseverelyunderstaffedrelativetoothermajor

patentofﬁces.GuellecandvanPottelsberghedelaPotterie(2007)

estimatethat“boththeincomingworkloadofexaminers(number

ofclaimsﬁledperexaminer)andtheiroutput(numberofclaims

grantedperexaminer)isthreetofourtimeshigherattheUSPTO

thanattheEPO”(p.201).ThusUSPTOexaminersspend

consid-erablylesstimethanEPOexaminersoneachclaim.Retentionof

experiencedpersonnelhasalsoprovedtobeparticularlydifﬁcult

attheUSPTO,duetoacombinationofrelativelylowsalariesand

attractiveoutsideoptions.5_Guellec_and_van_{Pottelsberghe}_de_la

Pot-terieconcludethatdecisionsattheEPOtendtobebasedonamore

carefulandexpertassessmentofthepriorartthanattheUSPTO,

whichmightexplainthemuchhigherrejectionratesinEurope.6

Second,USPTOexaminersappeartofacestrongerpressuresto

acceptratherthanrejectapplicationsthanexaminersinEuropeor

Japan.Onereasonisthat,bytransformingtheUSPTOina

user-fee funded entity, the 1990 Omnibus Act may have promoted

5_According_to_Mejer_and_van_{Pottelsberghe}_de_la_Potterie_(2011),_turnover_of employeesattheUSPTOisabout33%,comparedtojust3%attheEPO.

6_Quillen_and_Webster₍₂₀₀₁₎_estimate_that_the_patent_approval_rate_for_the_USPTO whencorrectedforcontinuingapplicationsisintherangeof95–97%,substantially higherthanfortheEPOorJPO.JaffeandLerner(2004)alsoshowthatsuccessful USPTOapplicationshaverisenintheperiodbetween1987and1998twiceasmuch as‘important’inventionsthatweregrantedinallthreeoftheworld’smajorpatent jurisdictions(theUSPTO,EPO,andJPO).Theyarguethatthisﬁndingis“hardto explaininanymannerotherthandecliningstandardsintheU.S.PTO,producing anever-growingproportionofU.S.patentsthepatent-holdersthemselvesdidnot thinkmeritedpatentingelsewhere”(p.143).Finally,Gallini(2002)notesthatifthe timeseriesofKortumandLerner(1999)isextendedtoincludethesecondhalfofthe 1990s,KortumandLerner’sconclusionthattheU.S.didnotbecomeanincreasingly attractivedestinationforforeigninventorsistempered,thus“lendingsomesupport forthe‘friendlycourt’hypothesis”(p.138).

(3)

1498 S.Belenzon,A.Patacconi/ResearchPolicy42 (2013) 1496–1510

a“customer-friendly”mentality.JaffeandLerner(2004)provide

several examples illustrating the dysfunctional effect of such

a mentality. Most importantly,USPTO examiners are rewarded

accordingtohowfasttheyprocessapplications.However,since

applicantscanmodifyandappealinitiallyrejectedpatents,

rejec-tionstendtobeverytime-consuming.Thus,USPTOexaminershave

strongﬁnancialincentivestoacceptratherthanrejectapplications.

Third,fromitsinceptiontheEPOallowsthirdpartiesto

chal-lengeclaimsinapatentforninemonthsafterthepatentisgranted.

Thispermitsadditionalpriorartreferencestobesubmittedand

allowsfurtherveriﬁcationofthepatentabilityconditions(Mejer

andvanPottelsberghedelaPotterie,2011).IntheU.S.,by

con-trast,asimilarpost-grantreviewprocesshasonlyrecentlybeen

introducedwiththe2011AmericaInventsAct.

Fourth, until quite recently, court decisions have tended to

weaken U.S. patentability standards. The CAFC established, for

instance, that prior publication obviates novelty only when all

thefeaturesofaninventionhavebeenrevealedbyasingleprior

publication,7_and_established_“the_reasonable_expectation_of

suc-cess”asalowerstandardofnon-obviousness8₍_Barton,_2000;_Lesser

andLybbert,2004).Therearealsofewerlimitationsonpatentable

subjectmattersintheU.S.thaninEurope.Inventionsintheareasof

software,businessmethodsandgenetics,forinstance,are

gener-allypatentableintheU.S.butnotinEurope.Theexplosionofpatent

activityinnewtechnologyareashasbeenviewedwithparticular

concernbycriticsoftheUSPTO.Examinationstandardsappearto

havebeentoolowintheseﬁelds(atleastinitially),dueto

inexpe-rienceonthepartofexaminers,lackofadequatewrittenpriorart

documents,andcourtdecisions(Merrilletal.,2004;Hall,2007).

Lastbutnotleast,patentingcostsaremuchlowerintheU.S.than

inEuropeorJapan.vanPottelsberghedelaPotterieandFranc¸ois

(2009)estimatethat anEPOpatentthatis renewedfor20 (10)

yearsin13memberstatescostsalmost9(4.6)timesasmuchasa

USPTOpatent.Anobviouseffectoflowpatentingcostsisto

encour-agelow-valueapplicationsandincreasethevarianceofthepatent

qualitydistributionbystretchingitsbottomtail.Perversefeedback

effectscanalsoarise,withlowpatentingcostsexacerbating

work-loadproblems.9 _Moreover,_ﬁrms_may,_when_they_have_a_choice,

startﬁlingstrategically,byﬁrstapplyingforUSPTOpatentsand

onlysubsequently,iftheinventionsturnouttobevaluable,forthe

moreexpensiveEPOpatents.Combined,theseeffectscouldlead

overtimetoawiderqualitygapbetweenEPOandUSPTOpatents.10

2.2. PatentportfolioevaluationinM&A:theeffectofinstitutional change

Fromaﬁnancialperspective,theﬁrmcanbeviewedasa

portfo-lioofinvestmentprojects;thusthevalueofthebusinessasawhole

canbecomputedasthenetpresentvalueofthecashﬂowsit

gener-ates.AsmentionedintheIntroduction,patents(andbyextension

patentportfolios)canbepositivelycorrelatedwithﬁrmvaluations

7_{Studiengesellschaft}_Kohle_v._Dart_Indus._Inc_.,₇₂₆_F.2d₇₂₄_(Fed._Cir._1984).

8_Amgen,_Inc_._v._Chugai_{Pharmaceutical}_Co_.,₉₂₇_F.2d₁₂₀₀_(Fed._Cir._1991).

9_How_changes_in_the_patent_fee_structure_can_improve_the_patent_{quantity-quality} tradeoffisanimportanttopicinpolicydebates,especiallyaftertheUSPTOraisedits feesin2004.SeeGuellecandvanPottelsberghedelaPotterieforadiscussion(2007, pp.212–213).

10_In_general,_ﬁrms_have_the_possibility_to_seek_patent_protection_in_multiple countries,butwillonlyusethispossibilityforinventionsthattheydeem sufﬁ-cientlyvaluabletojustifytheextraexpense.USPTOandEPOpatentsareinthis respectarguablydifferentbecause,whileaUSPTOpatentmayormaynotbepartof apatentfamilycomprisinganumberofequivalentpatentsindifferentcountries,an EPOpatentisalmostalwaystheresultofadecisionthattheinventionisworththe expenseofseekingprotectioninmultiplecountries,asasinglenationalapplication wouldotherwisebelessexpensive(wethankananonymousrefereeforpointing thisouttous).

because(i)theyestablishexclusionaryrightsonpatented

inven-tionsand(ii)provideusefulsignalsaboutthequalityofpatented

inventionsandoverallcorporatetechnologicalstrength.

Severalstudiesconﬁrmtheusefulness ofpatent statisticsas

indicatorsofﬁrms’inventiveeffortsandvalue(e.g.,Griliches,1990).

Dengetal.(1999)ﬁndthatthenumberofpatentsgrantedtoa

ﬁrminagivenyearandpatentcitationmeasures(forward

cita-tionsand“sciencelink”)arestronglyassociatedwithsubsequent

market-to-book ratios.Positiveassociations betweenﬁrmvalue

and citation-weighed patent indexes are also found by Bloom

andVan Reenen(2002),Halletal. (2005)and Belenzon(2012),

among many others. Pakes (1985) demonstrates that an

unex-pected increase in the number of patents granted toa ﬁrm is

associatedwithalargechangeinmarketvalue.Usinga

comprehen-sivesurveyinstrument,Gambardellaetal.(2008)directlyaskfor

informationabouttheperceivedvalueofpatents,ratherthan

try-ingtoinferpatentvaluesusingestimationtechniques.Theyreport

veryhighestimates,withtheaveragepatentvalueaboveEUR10

million,andamedianofEUR650thousand.Lastly,inasampleof

pharmaceuticalﬁrms,Narinetal.(1987)ﬁndthatpatentcountsare

highlycorrelatedwithseveralmeasuresofcorporatetechnological

strengthsuchasthenumberofnewdrugsregistered,important

newdrugsapproved,scientiﬁcarticlespublished,andexpert

opin-ionsofcorporatetechnicalstrength.

Inthispaper,wefocusonthecontributionofpatentportfolios

tothevaluationoftargetﬁrmsinM&Adeals.Inevaluatingpatent

portfolios,acquiringﬁrmsmusttakeseveralfactorsintoaccount.

Factorsfrequentlymentionedbypractitionersinclude:(i)changing

marketconditionsandtechnologycycles,(ii)thelevelofnoveltyof

thepatentedinventions,(iii)thestrengthandbreadthof

exclu-sionaryrights,(iv)thedifﬁcultyofinventingaround,(v)therisks

ofpatentdisclosure,(vi)aﬁrm’soverallportfolioposition

(includ-ingthepotentialforfurtherpatents),and(vii)bargainingpotential

(Wilson,2007).Moreover,inM&A thevalueofpotential

syner-giesbetweenacquiringandtargetﬁrmscanbepartlyincorporated

intothetargetﬁrm’sprice(SluskyandCaves,1991).Thus,patent

portfoliovaluationsinM&Amayalsoreflectfirm-specificand

deal-speciﬁcfactors,notjustthevalueofthepatentportfolioinsome

absolutesense.

The previous discussion suggests that patent reforms have

affectedboththelevelofnoveltyofthepatentedinventionsand

thestrength and breadthofexclusionaryrights (points(ii)and

(iii)intheclassiﬁcationabove).Weputforwardtwopotentially

conﬂictinghypotheses.Theﬁrstonestatesthat,becausepatents

havebecomemorepotentlegalweapons,theeffectofbothUSPTO

andEPOpatentsoncorporatevaluationshasgrownstrongerin

recenttimes.

H1(Strengtheningofpatentprotectionrights). Therelationship

betweenﬁrmvalueandUSPTOandEPOpatentshasgrownstronger

inrecenttimes.

The second hypothesis states that, because patent quality

appearstohave declinedmore in theU.S.than in Europe,EPO

patentsmayhavebecomeovertimeamorereliableindicatorof

technologicalstrengththanUSPTOpatents.Thus,theeffectofEPO

patentsonacquisitionvalueinM&Ashouldhavebecomestronger

overtime,relativetotheeffectofUSPTOpatents.Notethatitis

pos-siblethatthequalityofUSPTOpatentsmayhavedeclinedsomuch

astocompletelyoutweighthepotentialbeneﬁtsofstrongerpatent

protectionrights.Thus,incontrasttowhatsuggestedin

Hypothe-sis1,therelationshipbetweenUSPTOpatentsandﬁrmvaluemay

havegrownweaker,notstronger,overtime.

H2(DecliningUSPTOpatentquality). Overtime,EPOpatents

have become more strongly related to ﬁrm value than USPTO

(4)

Twoadditionalpointsareworthemphasizing.First,asdiscussed

above,patentqualityintheU.S.mayhavedeclinedmoreinnew

technologyareas suchas biotechnology, software and business

methodsthaninmoretraditionalareas(Barton,2000;Hall,2007).

Thus,changesintheimportanceofUSPTOpatentsovertimemay

notbeuniformacrosssectors.Wewilldistinguishbetween

differ-enttechnologysectorsintheempiricalpart.

Second,because theeconomic signiﬁcance of patents varies

tremendously,simplepatentcountsmaybeverynoisyproxiesfor

thevalueofaﬁrm’stechnology.Intheempiricalpart,wewilluse

forwardcitationstocontrolforthequalityofpatentedinventions.11

BecauseourdiscussionsuggeststhatUSPTOpatentsaremore

het-erogeneousintheirqualitythanEPOpatents(and,inparticular,

thattheyexhibitalongertailatthelowerendofthequality

distri-bution),weexpectcitation-weighingtostrengthenespeciallythe

effectofUSPTOpatents.

3. Data

Thispapercombinesdatafromthreesources:(i)M&Adatafrom

ThomsonSDCPlatinum,(ii)informationonpatentsfromtheUnited

StatesPatentandTrademarkOfﬁce(USPTO),and(iii)information

onpatentsfromtheEuropeanPatentOfﬁce(EPO).

WedropM&Adealsthatdonotreportinformationonvalue,net

totalassetsandacquiredstakes.Wealsodropdealswherethe

tar-getﬁrmisnotfromanOECDcountry,becausepatentlawmayvary

substantiallybetweendevelopedanddeveloping countries.This

leavesuswith30,306deals.40%ofthetargetsareAmerican,20%

areBritish,7%areJapanese,andtheremainingarefromotherWest

Europeancountries.Wematchtheseﬁrmstoourpatentdatasets.

Priortotheacquisitioncompletionyear,4,643ﬁrmshaveatleast

onepatentintheUSPTOorEPO.

Table1summarizesdescriptivestatisticsforthemainvariables

usedinouranalysis.Firmvalueiscomputedastheratiobetween

transactionvalueandacquiredstakes.Theaveragetargetﬁrmis

valuedat$228million,has$118millionintotalassets,generates

$138millioninannualsales,andmakes$27millioninproﬁts.Two

separatepatentstocksareconstructed:oneforUSPTOpatentsand

anotherforEPOpatents.Weincludeonlypatentsthataregranted

priortotheacquisitioncompletionyear.Stocksarecomputedusing

theperpetual inventorymethodwitha depreciation rateof 15

percent.Tocontrolforpatentquality,wealsoconstruct

citation-weighedstocksbyweighingeachpatentbythenumberofcitations

itreceives(excludingself-citations)dividedbytheaverage

num-berofcitationsreceivedbyallotherpatentsthataregrantedbythe

samepatentofﬁceinthesameyear.Foreachpatent,weinclude

onlycitations fromotherpatentsthat aregranted bythesame

patentofﬁce.Thatis,forUSPTOpatents,weonlyincludecitations

fromUSPTOpatents,andforEPOpatentsweonlyincludecitations

fromEPOpatents.Finally,weconstructanaggregatepatent

mea-sure–totalpatentstock–whichissimplythesumofUSPTOand

EPOpatentstocksforeachfirm.Theaveragepatentingfirm(afirm

withatleastonepatentfromtheUSPTOorEPO)hasastockof14

USPTOpatents(amedianof3),and8EPOpatents(amedianof1).

11 _Since_the_landmark_work_of_Trajtenberg_(1990),_{citation-weighed}_patent_indexes havebeenshowntobestronglycorrelatedwiththesocialvalueofinnovations (Trajtenberg,1990),peerevaluationoftheirtechnicalimportance(Albertetal., 1991),renewal decisions(Harhoffetal.,1999;Thomas,1999),andfirmvalue (Belenzon,2012;BloomandVanReenen,2002;Dengetal.,1999;Halletal.,2005; HirscheyandRichardson,2004).Halletal.(2005),inparticular,findthatinvestors areabletoaccuratelyforecasttheexpectedvalueofpatentedinventions,asitis laterconfirmedbyfuturecitations.Analternativewaytocontrolforpatentquality wouldhavebeentoestimatetheprivatevalueofpatentsusingannualrenewalfees, asinSchankermanandPakes(1986).

Since ourfocus is onhow therelationship between

patent-ingandacquisitionvaluehaschangedovertime,Table2presents

descriptivestatisticsacrosstimeperiods.Thesampleincludesboth

patentingandnon-patentingtargetﬁrms.Wesplitthesampleinto

sixtimeperiods.Averageﬁrmvalueisquitestableovertime.

Aver-agevalueis$233millionfor thedealscompletedintheperiod

1985–1990,and$249millionforthedealscompletedintheperiod

2005–2007.Ontheotherhand,thereisasubstantialriseinTobin’s

q–theratiobetweenvalueandassets–overtime.AverageTobin’s

qis3.4fordealscompletedin1985–1990,andis4.9fordeals

com-pletedin 2005–2007.Thus intangibleassetssuchasknowledge

appeartohavebecomemoreimportantovertime.

Table3examineshowkeycharacteristicsoftargetﬁrmshave

changedovertime.Onaverage,92%ofpatentingﬁrmshaveatleast

oneUSPTOpatent,and72%haveatleastoneEPOpatent.Overtime,

theshareofﬁrmsthatpatentintheUSPTOdeclinesbutremains

veryhighwhiletheshareofﬁrmsthatpatentintheEPOincreases

substantiallyfrom56%in1985toabout77%in2007.Thereisa

sharpincreaseinacquisitionvalueintheperiod1995–1998,but

thesehighvaluessubsequentlyreturntotheoverallsamplemean.

Thus,thereisnoclearpatternregardinghowﬁrmvalueinthese

acquisitiondealshaschangedovertime. Tobin’sq,ontheother

hand,increasessubstantiallyovertime,albeitinanon-monotonic

way(italsopeaksin1998).Theshareofnon-Americanacquisitions

alsochangessubstantiallyovertime(notreportedinatable).Inthe

wholesample40%oftheacquisitionsinvolveAmericantargets,yet

thisﬁguresfallsfromaround69%in1985toaboutaquarterofthe

dealsin2007.Wediscusshowthisissuemayaffectthe

interpre-tationofourﬁndinginthenextsectionandwhenpresentingthe

estimationresults.

Afactthatemergesclearlyfromourstatisticsisthatvaluations

wereveryhighduringthe1995–1998period.Themostlikely

expla-nationforthesehighvaluationsistheexuberancethataffected

investorsandstockmarketsduringthedot-combubbleyears.This

bubblestartedinMarch1995withtheexcitementoccasionedby

Netscape’sinitialpublicoffering,and terminated in 2000–2001

withthecollapseoftheNASDAQCompositeindex(however,the

Americaneconomystartedlosingspeedsince1999).Toaccountfor

possiblyinﬂatedcorporatevaluationsduringthisperiod,intherest

oftheanalysis(asinTable1),wewillreportseparateestimations

fortheperiods1995–1998and1999–2001,whichcorrespondto

theboomandbustofthedot-combubble.

4. Econometricanalysis

Weestimatethefollowingspeciﬁcationforthevalueoftarget

ﬁrms,bothforthewholesampleandfordifferenttimeperiods.12

lnValueit =ˇ0lnAssetsit+ˇ1ln(1+PatUSit−1)

+ˇ2ln(1+PatEUit−1)+t+i+cc+εi. (1)

Valueitis thevalueof targetﬁrmiatacquisitioncompletion

yeartandiscomputedastransactionvalueoveracquiredequity.

Assetsitistotalnetassets.PatUSit−1andPatEUit−1are,respectively,

targetﬁrmi’sUSPTOandEPOpatentstockspriortothe

acquisi-tioncompletionyear.t,iandccarecompletesetsofdummiesfor

acquisitioncompletionyears,twodigittargetindustrySICcodes,

andtargetcountry.εiisaniiderrorterm.Ourmaininterestisthe

waythecoefﬁcientsˇ1andˇ2changeovertime.Weinvestigate

thisintwoways:byinteractingUSPTOandEPOpatentstockswith

perioddummies,andbyestimatingseparatespeciﬁcationsfor

dif-ferenttimeperiods.BecausechangesinthequalityofUSPTOand

(5)

Table1

Summarystatisticsformainvariables.

Variable #acquisitions Mean Std.dev. Distribution

10th 50th 90th Targetvalue($,mm) 30,306 228 1,102 5 60 576 Netassets($,mm) 30,306 118 1,890 7 30 281 Tobin’sq($,mm) 30,306 4.0 4.9 0.7 2.1 10.2 Sales($,mm) 26,245 138 197 4 52 416 Proﬁts($,mm) 21,714 27 64 −2 7 79 %ofshareacquired 30,306 0.58 0.42 0.04 0.71 1

Totalpatentstock,count 4,643 22 194 0.7 4 42

Totalpatentstock,cite 4,643 21 161 0.4 4 43

USPTOpatentstock,count 4,643 14.4 142.9 0 2.6 25.5

USPTOpatentstock,citation-weighed 4,643 13.9 98.6 0.2 2.0 27.9

EPOpatentstock,count 4,643 7.7 54.3 0.5 1.1 14.5

EPOpatentstock,citation-weighed 4,643 7.5 64.8 0.3 0.9 13.5

Notes:Thistablereportssummarystatisticsforthemainvariableusedintheestimation.ThesampleincludesM&AtransactionsfromSDCPlatinumbetweenintheperiod 1985–2007.Firmvalueistheratiobetweentotaldealvalueandacquiredequity.Tobin’sqistheratiobetweenfirmvalueandassets.Patentvaluesarereportedforfirms withatleastoneUSPTOorEPOpatent.Weweighpatentsbycitationsbydividingthenumberofcitationstheyreceivebytheaveragenumberofcitationsreceivedbyall patentsthataregrantedbythesamepatentofficeinthesameyear.Robuststandarderrorsareinbrackets.

Table2

Summarystatisticsformainvariablesfordifferenttimeperiods.

Variable 1985–1990 (4,770deals) 1991–1994 (6,287deals) 1995–1998 (6,287deals) 1999–2001 (4,890deals) 2002–2004 (2,694deals) 2005–2007 (5,420deals) Mean Median Mean Median Mean Median Mean Median Mean Median Mean Median

Firmvalue($,mm) 233 72 194 37 222 64 262 73 204 56 249 74 Netassets($,mm) 108 37 154 19 92 31 103 33 110 38 126 29 Tobin’sq($,mm) 3.4 2.1 3.7 1.9 3.9 2.1 4.4 2.1 3.0 1.5 4.9 2.7 Sales($,mm) 167 74 123 40 120 42 143 55 146.4 60.3 141.8 58.2 Proﬁts($,mm) 42 15 23 5 27 7 27 8 22.9 5.2 23.5 5.4 %ofshareacquired 42.9 78.1 61.3 56.3 54.7 55.3 1.0 65.7 56.5 51.3 58.2 62.3

USPTOpatentstock,count 18.5 3.9 12.4 3.1 10.5 2.7 11.9 2.0 29.4 1.9 9.8 1.6

USPTOpatentstock,citation-weighed 18.3 3.8 14.0 2.5 13.0 2.2 11.0 1.4 21.6 1.1 8.0 1.1

EPOpatentstock,count 7.3 0.4 8.1 1.4 6.0 1.2 7.5 1.0 12.0 1.2 7.5 1.3

EPOpatentstock,citation-weighed 6.9 0.3 7.6 1.1 5.5 1.0 7.1 0.8 13.5 1.0 7.4 1.2

Notes:Thistablereportssummarystatisticsforthemainvariableusedintheestimationbymainperiods.ThesampleincludesM&AtransactionsfromSDCPlatinumbetween intheperiod1985–2007.Firmvalueistheratiobetweentotaldealvalueandacquiredequity.PatentdatafromtheUSPTOcovertheperiod1975–2007,andpatentsdata fromtheEPOcovertheperiod1979–2007.Weweightpatentsbycitationsbydividingthenumberofcitationstheyreceivebytheaveragenumberofcitationsreceivedby allpatentsgrantedinthesameyear.

Table3

Patentingﬁrmsacrossyears.

Year #deals Firmvalue Tobin’sq %PatentintheUSPTO %PatentintheEPO

1985 52 388.1 2.3 100.0 55.8 1986 98 510.8 3.0 98.0 62.2 1987 134 322.6 3.1 98.5 62.7 1988 187 319.1 2.4 95.2 62.0 1989 227 381.2 3.1 95.2 66.1 1990 216 393.4 3.1 94.0 72.2 1991 189 275.7 2.9 91.5 69.8 1992 186 278.8 3.0 94.1 75.8 1993 222 278.0 3.9 89.2 73.9 1994 280 254.9 3.6 94.3 71.4 1995 253 466.7 4.2 96.0 71.1 1996 282 312.6 3.9 94.3 69.5 1997 285 390.3 5.0 96.1 73.3 1998 86 639.5 4.3 94.2 67.4 1999 336 275.3 4.2 90.8 73.8 2000 211 306.2 4.1 92.4 73.5 2001 137 213.0 3.3 93.4 74.5 2002 229 296.4 2.9 90.8 73.8 2003 203 179.8 2.8 91.6 77.8 2004 30 598.9 4.0 93.3 73.3 2005 246 308.0 4.6 85.0 76.8 2006 357 342.7 3.9 82.6 78.2 2007 197 358.1 3.6 83.2 76.6 Total 4,643 329.8 3.6 91.9 72.1

Notes:Thistablereportsthenumberofpatentingfirmsforeachyearinoursample.WeincludeonlyfirmsthathaveatleastonepatentinUSPTOorEPObytheiracquisition year.Firmvalueistheratiobetweentotaldealvalueandacquiredequity.Tobin’sqistheratiobetweenfirmvalueandassets.%patentintheUSPTO(EPO)isthepercentage offirmsthathaveatleastonepatentintheUSPTO(EPO)outofallpatentingfirmsinthesameyear.

(6)

EPOpatentsmayinpartdriveourresults,wewillalsocomparethe estimatesˇ1andˇ2forspeciﬁcationswherepatentsareweighed bycitationsandspeciﬁcationswherepatentsarenotweighedby citations.13

Thereareanumberofimportantpotentialbiasesthatmayaffect

ourestimationswhichareworthdiscussingbeforepresentingthe

results.

Afeatureofoursampleisthatitsﬁrmcompositionchangesfrom

oneyeartoanother.Thisleavesourresultssensitivetounobserved

heterogeneitybias.Aconcernmightbe,forinstance,thatﬁrmsthat

choosetopatentintheEPOinthelatesampleperiodmaybe,on

average,ofhigher(unobserved)qualitythanﬁrmsthatpatentin

theEPOinthebeginningofthesampleperiod.Thismeansthat

laterinthesampleperiodthecorrelationbetweenEPOpatentsand

unobservedﬁrmqualityispositive,causinganupwardbiasinˇ2.

Wecancheckwhetherthisconcernislikelytoaffectourestimates

bycomparingobserveddealcharacteristicsforearlyandlaterEPO

patentees.ForﬁrmsthatpatentintheEPOintheﬁrstestimation

period,1985–1990,averagedealvalueis$395million,Tobin’sqis

2.9,salesare$242million,andproﬁtsare$63million.Forﬁrmsthat

patentintheEPOinthelastsampleperiod(2005–2007),average

valueislowerat$370million,butTobin’sqishigherat4.2.Salesare

similartotheearlyperiodlevelat229,butproﬁtsaremuchlower

at$33million.Fromthiscomparison,itisnotclearwhetherearlier

EPOpatenteesareoflowerqualitythanlaterones.Interestingly,we

doﬁndthatthatlaterEPOpatenteesbeneﬁtfromhigherTobin’sqas

comparedtoearlyEPOpatentees.Thisisconsistentwiththeview

thatknowledgeassetsfortheseﬁrmsaccountforagreaterportion

ofﬁrmvalue.Ourestimationresultswhicharepresentedbeloware

consistentwiththisinterpretation.

Anotherpotentialsourceofsampleselectionisindustry

com-position. In the later period of the sample we observe more

biotechnologyandinformationtechnologyacquisitionsthanin

ear-lierperiods.Iftheimpactofpatentsonﬁrmvaluationvariesacross

industries, then ourkey estimates could bebiased. Controlling

forindustryisalsoimportantbecause,asourdiscussionsuggests,

patentqualitymayhavedeclinedmoreinsomeindustriesthan

others.Tocheckthesensitivityofourresultstoindustry

varia-tion,wereportourmainestimationresultsseparatelyforthemain

innovativeindustriesinoursample.14

Lastly,weobserveasharpdeclineintheshareofAmerican

tar-getsinoursample.Whileintheearlyperiodsabout50percentof

targetsareAmerican,thispercentagedropstoabout20percentin

thelatesampleperiods.Asimilarpatternemergeswhenlookingat

patentingﬁrmsonly(from60to30percent).Thiscouldaffectour

results,forinstanceifUSPTOpatentsmattermostlytoAmerican

ﬁrms,butalsonon-AmericanﬁrmspatentattheUSPTO.Thus,when

presentingtheeconometricresults, wealsodiscuss the

robust-nessofourﬁndingstosplittingthesamplebetweenAmericanand

non-Americantargets.

13 _An_important_caveat_is_in_order._A_premise_of_our_analysis_is_that_patent_citations areausefulproxyforpatentquality.However,becausethecitationsapatentreceives accumulateovertime,patentqualityislikelytobemeasuredwitherrortowardthe endoftheestimationperiod.IftheEPOprovidessomequalityreassurancebecause few“bad”patentsaregrantedbytheEPO,thenˇ2islikelytobehigher(andˇ1lower) inthelastyearsoftheestimationperiodevenifpatentsareweightedbycitations, simplybecauseofcitation-truncation.Yet,ﬁndingthatUSPTOpatentsbecomeless valuablebeforetheendoftheestimationperiod(aswedo,seeTable5,columns4–5 and10–11)suggeststhattruncationisnotlikelytocompletelydriveourresults.

14 _We_also _checked _how _the _{representation} _of _different _industries _changes throughoutoursampleperiod.Wefocusedondrugsandchemicalsbecause patent-ingisveryimportantintheseindustries;thuschangesinrepresentationovertime havethepotentialtoaffectourestimates.Wedidnotﬁndsubstantialvariationin therepresentationoftheseindustriesinoursampleovertime.In1985,4.6%ofthe dealsareindrugsandchemicals.In2007,theproportionis5.0%.

5. Estimationresults 5.1. “Horse-race”estimation

Table4reportstheestimationresultsforthecompletesample

period.Westartwithpatentmeasuresnotweighedbycitations

(thus,weusethenumberofpatentseachﬁrmholds when

con-structingtheEPOandUSPTOpatentstocks).Column1includes

loggedtotalpatentstock(thelogofthesumofUSPTOandEPO

patentstocks). Theelasticityestimateof acquisitionvalue with

respecttototalpatentstockis0.117(astandarderrorof0.007).

Obtaininganelasticityestimateallowsustocalculatetheincrease

inacquisitionvalueasaresponsetoaunitaryincreaseinpatent

stock.Forthiscomputation,weusetheaverageacquisitionvalue

andaveragepatentstockvaluesforpatentingﬁrmsinthesample.

Theelasticityestimatefromcolumn1impliesthatincreasingtotal

patentstockbyoneisassociatedwitha$1.2millionincreasein

acquisitionvalue.

Columns2and3addthestocksofUSPTOandEPOpatents

sep-arately.Inbothcasesthecoefﬁcientestimatesarelargeandhighly

signiﬁcant.Incolumn4weestimatea“horse-race”betweenUSPTO

andEPOpatents.Includingbothmeasuresinasingleregression

resultsinadropinbothcoefﬁcientestimatesascomparedtowhen

eachvariableisseparatelyincluded.Whenbothstocksareincluded,

theelasticityofacquisitionvaluewithrespecttoEPOpatentsis

almosttwiceaslargeastheelasticitywithrespecttoUSPTOpatents.

Basedontheestimatesfromcolumn4,raisingtheEPOpatentstock

byoneunitisassociatedwitha$2.6millionincreaseinﬁrmvalue,

comparedtoa$0.8millionincreasefortheUSPTOpatentstock.

Columns 5–7 report initialrobustness checks for the above

results.Incolumn5weexcludedealsininformationtechnology

andbusinessservicesbecausepatentabilityrequirementsinthese

ﬁeldsvarysubstantially betweentheU.S.and Europe.The

esti-matesarenotsensitivetoremovingtheseindustries.Incolumn

6weincludeonlypatentingﬁrmstoensurethatourresultsarenot

drivenbycomparingpatentingtonon-patentingﬁrms.Ifselection

intopatentingiscorrelatedwithanomittedvariablethatisrelated

tovalue,thenourpatentingvariableswouldbeupward-biased.We

ﬁndthesamepatternofresultswhenfocusingonlyonﬁrmswith

atleastonepatent.Incolumn7weincludeonlydealsthatare

clas-siﬁedas“high-tech”inSDC.Ourconcernhereisthatourestimates

maybedrivenbycomparinghigh-techdeals(wherepatent

valua-tionsmaybehigh)tolow-techdeals(wherepatentvaluationsmay

below).Theestimatesremainstable.

Columns8–14presenttheestimationresultswhencontrolling

forpatentqualityusingforwardcitations.Controllingforcitations

proves importantfor USPTO patents,but much less sofor EPO

patents.ThisisexpectedundertheassumptionthatUSPTOpatents

areofgreaterqualityheterogeneitythanEPOpatents.Controlling

forcitationsincreasesthecoefﬁcientestimateforUSPTOpatents

bynearly30%(from0.056to0.072).Ourestimatesimplythata

unitaryincreaseincitation-weighedEPOandUSPTOpatentstocks

is associated with$2.3and $1.1 millionincrease in acquisition

value,respectively.These estimatesstill implylargedifferences

inthevalueofUSPTOandEPOpatents;however,thedifference

isconsiderablylowerthanwhenwedonotcontrolforcitations.

AnotherinterestingresultistherobustnessoftheUSPTO

coefﬁ-cientincolumns12–14wherewe performthesamesensitivity

checksasreportedincolumns5–7.

Comparingourﬁndingstopreviousestimatesischallengingdue

totheheterogeneityof methodsused andthefact that weare

notawareofotherstudiesthatfocusonacquisitionvalue.These

caveatsnotwithstanding,weprovideafewcomparisonsto

lead-ingstudiesintheﬁeld.First,we compareourestimatestoHall

et al.(2005).Like us, theyuseﬁrm valuetoinfer patentvalue

(7)

1502 S. Belenzon, A. Patacconi / Research Policy 42 (2013) 1496 – 1510 Table4

Patentsandﬁrmvalue.

Dependentvariable:ln(Firmvalue)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

Patentcount(notweighedbycitations) Patentsweighedbycitations

All All All All Exc.ITﬁrms Atleastonepatent OnlyHigh-Tech All All All All Exc.ITﬁrms Atleastonepatent OnlyHigh-Tech ln(1+TotalPatentStock)t−1 0.117** 0.122**

(0.007) (0.007) ln(1+USPTOPatent Stock)t−1 0.117** _0.056** _0.056** _0.048** _0.050** _0.125** _0.075** _0.076** _0.072** _0.072** (0.008) (0.012) (0.012) (0.015) (0.012) (0.008) (0.011) (0.012) (0.012) (0.012) ln(1+EPOPatent Stock)t−1 0.148** _0.100** _0.093** _0.090** _0.097** _0.152** _0.086** _0.079** _0.080** _0.080** (0.010) (0.014) (0.014) (0.014) (0.014) (0.010) (0.014) (0.014) (0.014) (0.014) ln(Netassets) 0.713** _0.714** _0.715** _0.713** _0.727** _0.716** _0.710** _0.712** _0.714** _0.715** _0.713** _0.727** _0.709** _0.709** (0.004) (0.004) (0.004) (0.004) (0.005) (0.012) (0.005) (0.004) (0.004) (0.004) (0.004) (0.005) (0.005) (0.005)

Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

R2 _0.680 _0.679 _0.679 _0.679 _0.689 _0.683 _0.653 _0.680 _0.679 _0.679 _0.680 _0.690 _0.653 _0.653

Observations 30,306 30,306 30,306 30,306 26,532 4,643 24,010 30,306 30,306 30,306 30,306 26,532 24,010 24,010

Notes:ThistablereportstheresultsofOLSregressionsthatexaminetherelationshipbetweenpatentstocksandﬁrmvalue.ThesampleincludesM&AtransactionsfromSDCPlatinumbetweenintheperiod1985–2007.Patent

datafromtheUSPTOcoverstheperiod1975–2007,andpatentsdatafromtheEPOcoverstheperiod1979–2007.InColumns8–14weweighpatentsbycitations.Wedividethenumberofcitationstheyreceivebytheaverage

numberofcitationsreceivedbyallpatentsgrantedbythesamepatentofficeinthesameyear.Columns5and12excludefirmstheoperateinfieldsofinformationtechnologiesorbusinessservices.Robuststandarderrorsarein

brackets.

*_Signiﬁcant_at_5%.

(8)

data).Theyreportacitationselasticityofabout0.03,whileweﬁnd

amuchhigher(citation-weighed)elasticityof0.122.Itis

impor-tanttonote,however,thatHalletal.(2005)alsocontrolforR&D

–informationwhichwedonothave.Thus,ourestimated

elastic-ityislikelytobepickingtheR&Deffectaswell(forinstance,an

increaseinnon-patentedknowledge).Interestingly,ourestimates

becomemuchclosertothoseofHalletal.(2005)when

consider-ingthehighly-citedpatentsintheirsample,andactuallybecome

lowerthantheirestimatesforﬁrmswithextremelylarge

num-bersofcitationsperpatent.Ourestimatesarealsomuchhigher

thatpreviousestimatesobtainedfrompatentrenewaldata.For

example,usingrenewaldataonEuropeanpatents,Pakes(1986)

andSchankerman andPakes (1986)reportvery lowvalue

esti-matesrangingfrom$847to$19,124(in1980prices).However,the

renewalapproachonlycapturesthe“patentpremium”–the

incre-menttothevalueofaninnovationwhichisrealizedbypatenting

it–nottheoverallvalueofapatentedinvention.Thelatteriswhat

wemeasure.Estimatesusingrenewaldatamayalsobedownward

biasedbecauseextremelyvaluablepatentsmaynotbeproperly

accountedfor(Harhoffetal.,2003;Aroraetal.,2008).Ourestimates

aremuchclosertothoseobtainedbyGambardellaetal.(2008)

usingthecomprehensivePATVALsurveyonthevalueofEuropean

patents.ThePATVALsurveydirectlyaskedforinformationofthe

perceivedvalueofpatents,ratherthantryingtoinfervalueusing

differentestimationtechniques.Theauthorsreportveryhighvalue

estimates,withaveragepatentvalueofaboveEUR10million,and

amedianofEU650thousand.Ourestimatesarequiteclosetotheir

surveymedianvalue.

5.2. Variationovertime

Next,westudyhowthevalueofUSPTOandEPOpatentshas

changedovertime.Forthispurposeweestimateseveral

“horse-race”speciﬁcationsfordifferentperiodsinoursample.Thetime

periodsareconstructedtryingtoeventhedistributionofnumberof

dealsovertime,whilealsoisolatingsomespeciﬁchistoricalfactors

(thedot-combubblein1995–1998anditsburstin2000–2001)that

areevidentinourdescriptivestatistics.

Table 5 presents the estimation results. Two speciﬁcations

areprovided: ﬂexible speciﬁcationswhere we splitthe sample

byperiods, and speciﬁcations where we include an exhaustive

setofinteractionsbetweentheperioddummiesandUSPTOand

EPO patent stocks. We begin with the case when patents are

notweighedbycitations(columns1–6).Aclearpatternemerges.

USPTOpatentsareimportantintheﬁrstsampleperiod1985–1990,

butsubsequentlyhavenosigniﬁcanteffectonacquisitionvalue,

exceptthaninthedot-combubbleyears1995–1998.Bycontrast,

theeffectofEPOpatentsisalwayslargeandsigniﬁcantfrom1991

to2007(columns1–5).

Estimationusingapooledregression(column6)conﬁrmsthese

results.EPOpatentshaveastableeffectonvalue:wecannotreject

thehypothesisthattheEPOperiodinteractionsaredifferentfrom

oneanother(p-value=0.82).Yet,wecanstronglyrejectthe

hypoth-esisthattheyarejointly equaltozero(p-value<0.001). Onthe

otherhand,USPTOpatentsdonothavearobusteffecton

acqui-sitionvalue, andstarting from1991, withtheexception ofthe

1995–1998period,theireffectonacquisitionvalueisstatistically

zero.TheimportanceofEPOpatents,relative toUSPTOpatents,

clearlyincreasesovertime.Intheﬁrstsampleperiod,the

differ-encebetweenEPOandUSPTOcoefﬁcientestimatesis0.017.The

samedifferenceintheﬁnalsampleperiodis0.120.

Asimilarpatternemergeswhenpatentsareweighedby

cita-tions(columns7–12).USPTOpatentsmatterforvalueintheﬁrst

three sample periods (1985–1998), but their effect completely

disappears in the 1999–2007 period. In contrast, EPO patents,

whichdonothaveasigniﬁcanteffectonvalueintheﬁrstperiod

0.16 0.18 0.2 0.08 0.1 0.12 0.14 0.16 EPO average 0.39 0.45 0.43 0.43 0.32 0.46 0 0.02 0.04 0.06 USPTO average 0.2 1985-1990 1991-1994 1995-1998 1999-2001 2002-2004 2005-2007 Share number of EPO patents USPTO patents EPO patents

a

b

0.12 0.14 0.16 0.18 0.2 0 37 0.60 0.44 _0.41 0.50 0.04 0.06 0.08 0.1 USPTO average EPO average 0.37 0.35 0.41 0 0.02 1985-1990 1991-1994 1995-1998 1999-2001 2002-2004 2005-2007 Share number of EPO patents USPTO patents EPO patents Fig.1.(a)Time-varyingcoefﬁcientestimatesonUSPTOandEPOpatentstocks. Patentsarenoweighedbycitations.(b)Time-varyingcoefﬁcientestimatesonUSPTO andEPOpatentstocks.Patentsarenotweighedbycitations.

(1985–1990),becomethedominanttechnologyindicatorinthelate

sampleperiods(1999–2007).15

Comparingweighedandun-weighedspeciﬁcationsshowsthat

citation-weighingbolsterstheeffectoftheUSPTOpatentsinthe

earlysampleperiods.Thatiswhatonewouldexpectifthe

qual-ityofUSPTOpatentsisveryheterogeneous.However,eveninthe

citation-weighedspeciﬁcations,thecoefﬁcientonUSPTOpatents

dropstozerointhelatesampleperiods.Apossibleexplanationis

thatbecausecitationsaccumulateovertime,patentqualitymaybe

measuredwithlargeerrorneartheendofthesampleperiod.Thus

citation-weighingmaynotbeaparticularlygoodwaytocontrol

forpatentqualityinthelastfewyearsofoursample.Theeffectof

EPOpatents,bycontrast,alwaysremainsverystrong,bothinthe

un-weighedandweighedregressions.Fig.1graphicallyillustrates

theresultsfromcolumns6and12.

Overall,theresultsdonotsupportthehypothesisthatstronger

patentprotectionrightsledtohighervaluationsforUSPTOandEPO

patents(Hypothesis1).ThereissomeevidencethatEPOpatents

becamemorevaluableinthelasttwosampleperiods(2002–2004

and 2005–2007); however, USPTO patent coefﬁcients certainly

didnotrise.Theresultsappear tobemore consistentwiththe

15_Total_patents_stock_–_the_sum_of_USPTO_and_EPO_stocks_–_has_a_very_stable_effect overtime(notpresentedinthetable).Thecoefﬁcientestimateonthe (citation-weighed)totalpatentsstockintheperiod1985–1990is0.102,comparedtoavalue of0.122inthe2002–2007period.

(9)

1504 S. Belenzon, A. Patacconi / Research Policy 42 (2013) 1496 – 1510 Table5

Patentsandﬁrmvalueovertime.

(1) (2) (3) (4) (5) (5) (6) (7) (8) (9) (10) (11) (12) (13)

1985–1990 1991–1994 1995–1998 1999–2001 2002–2004 2005–2007 All 1985–1990 1991–1994 1995–1998 1999–2001 2002–2004 2005–2007 All ln(1+USPTOPatent Stock)t−1 0.073** _0.033 _0.103** _0.032 _0.032 _0.003 _0.084** _0.064** _0.125** _0.056 _0.041 _0.014 (0.018) (0.030) (0.030) (0.036) (0.052) (0.034) (0.017) (0.028) (0.030) (0.033) (0.047) (0.034) ln(1+EPOPatent Stock)t−1 0.056 0.125** _0.087** _0.107** _0.134* _0.149** _0.043 _0.103** _0.069* _0.094** _0.138* _0.143** (0.023) (0.033) (0.030) (0.041) (0.063) (0.039) (0.024) (0.033) (0.031) (0.038) (0.059) (0.039)

ln(1+USPTOPatentStock)t−1×

1985≤Year≤1990 0.046** _0.055** (0.017) (0.016) 1991≤Year≤1994 0.036 0.059* (0.030) (0.027) 1995≤Year≤1998 0.169** _0.185** (0.028) (0.028) 1999≤Year≤2001 0.053 0.099** (0.033) (0.031) 2002≤Year≤2004 −0.003 −0.000 (0.048) (0.044) 2005≤Year≤2007 −0.012 −0.000 (0.033) (0.033)

ln(1+EPOPatentsStock)t−1×

1985≤Year≤1990 0.063** _0.053* (0.024) (0.025) 1991≤Year≤1994 0.131** _0.114** (0.033) (0.032) 1995≤Year≤1998 0.086** _0.072** (0.030) (0.031) 1999≤Year≤2001 0.092* _0.061** (0.040) (0.038) 2002≤Year≤2004 0.139** _0.144** (0.059) (0.054) 2005≤Year≤2007 0.118** _0.110** (0.038) (0.037) ln(Netassets) 0.734** _0.711** _0.726** _0.711** _0.751** _0.674** _0.713** _0.734** _0.711** _0.725** _0.711** _0.751** _0.674** _0.713** (0.011) (0.009) (0.010) (0.010) (0.014) (0.010) (0.004) (0.011) (0.009) (0.010) (0.010) (0.014) (0.010) (0.004)

Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

R2 _0.726 _0.693 _0.686 _0.671 _0.683 _0.664 _0.680 _0.726 _0.693 _0.686 _0.671 _0.683 _0.664 _0.680

Observations 4,770 6,287 6,245 4,890 2,694 5,420 30,306 4,770 6,287 6,245 4,890 2,694 5,420 30,306

Notes:ThistablereportstheresultsofOLSregressionsthatexaminetherelationshipbetweenpatentsandﬁrmvalueindifferenttimeperiods.Robuststandarderrorsareinbrackets.

(10)

hypothesisthatUSPTOpatentqualityandvaluationsdeclinedover

time(Hypothesis2).

Usingtheestimatesfromcolumn13, wecanagaincalculate

patentvaluesfordifferenttimeperiodsseparatelyforUSPTOand

EPOpatents.Intheﬁrstestimationperiod,1985–1990,aunitary

increaseinUSPTOpatentstockleadstoa$765thousandincrease

inacquisitionvalue.Inthethirdperiod,1995–1998,whereUSPTO

patentsseemtobethemostvaluable,thesameunitaryincrease

leadstoa$2.8millionriseinacquisitionvalue.Thisvalue,

how-ever,dropstozerointhe2002–2007period.EPOpatentsexhibit

amuchmorestablecorrelationwithacquisitionvalue.Intheﬁrst

estimationperiod,1985–1990,aunitaryincreaseinthestockofEPO

patentsleadstoa$1.4millionincreaseinvalue.Thisvaluerisesto

$3.1millioninthesecondperiod1991–1994,whichisthesame

valuethatisobtainedinthelastestimationperiod(2005–2007).

Inunreportedspeciﬁcations,werepeatedtheanalysisinTable5

forpatentingfirmsonly,forfirmsclassifiedashigh-techbySDC,

andexcludingITandbusinessservices.Thesamepatternofresults

continuestohold.16

6. Robustness

Thepreviousanalysis uncoveredtwo important ﬁndings:(i)

overtime,EPOpatentsbecomemorevaluablerelativetoUSPTO

patents,and(ii)theeffectofUSPTOpatentsonacquisitionvalue

completelydisappearsinthelatesampleperiods.Inthissection

weexploretherobustnessoftheseﬁndingsbysplittingthesample

accordingtodifferentcriteria.Duetothefewerdegrees-of-freedom

associated with suchexperimentation, the sample is split into

fewertimeperiods.

6.1. Targetﬁrmsize

Patentsplayaparticularlyimportantroleinthestart-upphaseof

aﬁrm’slife(Long,2002).Theyprotectyoung,smallﬁrmsfrom

com-petitionbyindustryleaders(Teece,1986).Topotentialinvestors,

they provide “evidence that the company is well managed, is

ata certain stage in development,and hasdeﬁned and carved

outamarketniche”(Lemley,2001,pp.1505–1506).Not

surpris-ingly,therefore,venturecapitalists,businessangelsandcorporate

investorsincreasinglybasetheirinvestmentdecisionsonIP

con-siderations(KortumandLerner,2000).

Table6examinestheextenttowhichourresultsaredrivenby

differencesintargetﬁrmsize.Weclassifyaﬁrmassmallifitssales

fallinthelowestquartileofthesalesdistribution,andaslargeif

itssalesfallinthehighestquartile.Consistentwiththeideathat

patentsmatterthemostforsmallﬁrms,ourresultsholdonlyin

thesmall-firmsample.For largefirmswefindnoevidencethat

patents(eitherUSPTOorEPO)matterforvaluation,andthereisno

importanttemporalvariationintheirimportance.Bycontrast,EPO

patentsstronglycorrelatewiththevaluationofsmallﬁrms,

espe-ciallyinthelastsampleperiodwhentheyarealmostasimportant

forvaluationasnetassets.TheeffectofUSPTOpatentsonsmall

16 _We_also_more_closely_examine_the_rise_in_USPTO_patent_values_during_the_bubble years1995–1998.TheriseinUSPTOpatentvaluesisrobusttoremovinganyyear inthisperiod.Forinstance,whenremovingyear1998(anoutlierintermsofdeal valuesasshowninTable3),thecoefficientestimateonUSPTOpatents (citation-weighed)is0.125andthecoefficientestimateonEPOpatentsis0.063.Splittingthe samplebyAmericanandnon-AmericantargetsshowsthattheriseinUSPTOpatent valuesisdrivenbynon-U.S.targets.ForAmericantargets,theUSPTOcoefficient estimateis0.070;thesameestimatefornon-Americantargetsis0.227.Lastly,the coefficientestimateonUSPTOpatentsisabout50%largerinthe1995–1998period fortargetsintheinformationtechnologyfieldthaninalltheotherindustriesinthe

sameperiod. Table

6 Low sales vs. high sales firms. Dependent variable: ln( Firm value ) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Patent count (not weighed by citations) Patents weighed by citations Sales ≤ 25th percentile Sales > 75th percentile Sales ≤ 25th percentile Sales > 75th percentile All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 ln(1 + USPTO Patent Stock )t− 1 0.072 ** 0.136 ** 0.059 − 0.070 0.034 ** 0.044 ** 0.005 0.017 0.149 ** 0.202 ** 0.126 − 0.015 0.043 ** 0.045 ** 0.019 0.045 (0.047) (0.060) (0.133) (0.110) (0.017) (0.021) (0.050) (0.043) (0.040) (0.048) (0.104) (0.106) (0.016) (0.019) (0.048) (0.043) ln(1 + EPO Patent Stock )t− 1 0.305 ** 0.283 ** 0.103 0.489 ** 0.025 0.031 0.000 0.059 0.241 ** 0.205 ** 0.043 0.474 ** 0.022 0.037 − 0.009 0.039 (0.052) (0.062) (0.157) (0.123) (0.020) (0.024) (0.056) (0.046) (0.053) (0.062) (0.143) (0.133) (0.019) (0.023) (0.050) (0.045) ln( Net assets ) 0.539 ** 0.535 ** 0.536 ** 0.544 ** 0.613 ** 0.603 ** 0.551 ** 0.661 ** 0.738 ** 0.534 ** 0.533 ** 0.543 ** 0.612 ** 0.602 ** 0.550 ** 0.660 ** (0.012) (0.016) (0.033) (0.021) (0.018) (0.027) (0.034) (0.031) (0.012) (0.016) (0.032) (0.021) (0.018) (0.027) (0.034) (0.031) Two-digit industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country target dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Acquisition year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes R 2 0.497 0.521 0.519 0.521 0.433 0.409 0.507 0.525 0.498 0.521 0.520 0.522 0.434 0.409 0.507 0.525 Observations 6,562 3,940 1,058 1,564 6,561 3,619 1,134 1,808 6,562 3,940 1,058 1,564 6,561 3,619 1,134 1,808 Notes : This table reports the results of OLS regressions that examine the relationship between patents and firm value for different periods. We distinguish between firms with low and high revenues. Robust standard errors are in brackets. *Significant at 5%. ** Significant at 1%.

(11)

ﬁrms’acquisitionvalue,while strongintheﬁrstsampleperiod,

subsequentlydisappears.

6.2. Target-acquirersimilarity

Thissectionexplorestheextenttowhichourﬁndingsaredriven

bysimilaritiesbetweenacquiringandtargetﬁrms.Wefocusontwo

dimensionsofsimilarity:whethertheacquiringandtargetﬁrms

operateinthesametwo-digitSICcode(productmarket

similar-ity),andwhethertheacquiringandtargetﬁrmsoperateinthesame

country(geographicproximity).Weconjecturethat,iftheEPOlabel

providesastrongersignalofqualitythantheUSPTOlabel,thenEPO

patentsshouldbeparticularlyvaluable,relativetoUSPTOpatents,

inbetween-industryandbetween-countrydeals.Inthesedeals,in

fact,asymmetricinformationislikelytobeparticularlyprominent,

asacquiringﬁrmsmaylackspeciﬁcindustryknowledgeor

knowl-edgeaboutlocalconditions.Thus,thequalitycertiﬁcationprovided

bytheEPOlabelmaybeespeciallyvaluable(Akerlof,1970).

Tables 7and 8 focusonproduct marketsimilarity and

geo-graphicproximity, respectively. Table7 segregates within- and

between-industryacquisitionsusingthetwo-digitSICcodesofthe

acquiringand targetﬁrms.We ﬁndthatover timeEPOpatents

becomemoreimportant,relativetoUSPTOpatents,for

between-industryacquisitionsrelativetowithin-industryacquisitions.This

istrueregardlessofwhetherpatentsareweighed,althoughmuch

strongerUSPTOeffects areobtainedwhenpatentsareweighed.

Table8segregatesdomesticandcross-borderacquisitions.Here,

USPTOpatentsmatteronlyforvalueindomesticdealsandthe

ear-lierperiod.Bycontrast,EPOpatentshaveastrongeffectonvalue

bothin domesticand cross-borderacquisitions.TheriseinEPO

patentvalues,however,isparticularlypronouncedincross-border

acquisitions.

Overall,ourﬁndings indicatethatEPO patentshavebecome

especiallyimportantinsituationswhereasymmetricinformation

ismostlikelytobesubstantial(i.e.,between-industryand

cross-countrydeals).ThissuggeststhepossibilitythattheEPOlabelmay

providesomeformofqualityreassurance.Exploringthisissue

fur-therisaninterestingdirectionforfutureresearch.

6.3. Targetﬁrmnationality

Animportantconcern isthat ourtemporal patternsmaybe

drivenbychangesinsamplecomposition.Mostimportantly,the

proportionofEuropeantargetsinoursamplerisessubstantially

overtime.17_If_USPTO_patents_are_less_important_for_European_ﬁrms

becausetheymostlysellinEuropeanmarkets,thenapatternof

decliningUSPTOpatentvaluesmightbeobserved.Tomitigatethis

concern,Table9splitsthesamplebytargetﬁrmnationality(inthis

tablepatentsarealwaysweighedbycitations).Weﬁndthat,for

Americantargets,USPTOpatentshavealargeeffectonvalueinthe

earliersampleperiod,butthiseffectcompletelydisappearsbythe

endofthesampleperiod.EPOpatents,ontheotherhand,

experi-enceanoppositepattern:theyhavenoeffectonvalueintheearlier

sampleperiod,andalargeeffectinthelaterperiod.ForEuropean

targets,thepatternissimilar.USPTOpatentsareonlyimportant

forvalueintheearliersampleperiods,whileEPOpatentshavea

strongandsigniﬁcanteffectinbothearlyandlateperiods(with

muchstrongerlate-periodeffects).Thus,ourpatterns–andin

par-ticularthepatternofdecliningUSPTOeffects–donotseemtobe

17 _There_is_a_strong_{“home-bias”}_effect_in_{acquisitions.}₈₄_percent_of_acquisitions byAmericanﬁrmsareAmerican,andalmostallacquisitionsbyEuropeanﬁrmsare

European.Thesepatternsarestableovertime. Table

7 Within-vs. between-industry acquisitions. Dependent variable: ln( Firm value ) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Patent count (not weighed by citations) Patents weighed by citations Target and acquirer are in the same 2-digit SIC code Target and acquirer are in different 2-digit SIC codes Target and acquirer are in the same 2-digit SIC code Target and acquirer are in different 2-digit SIC codes All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 ln(1 + USPTO Patent Stock )t− 1 0.077 ** 0.080 ** 0.062 0.048 0.044 ** 0.051 ** 0.007 − 0.009 0.097 ** 0.100 ** 0.087 0.067 0.064 ** 0.072 ** 0.033 − 0.010 (0.019) (0.022) (0.060) (0.045) (0.015) (0.018) (0.045) (0.037) (0.018) (0.021) (0.052) (0.043) (0.015) (0.018) (0.045) (0.037) ln(1 + EPO Patent Stock )t− 1 0.071 ** 0.084 ** 0.052 0.105 * 0.118 ** 0.105 ** 0.160 ** 0.163 ** 0.054 ** 0.063 ** 0.033 0.092 0.107 ** 0.090 ** 0.152 ** 0.169 ** (0.022) (0.024) (0.069) (0.051) (0.018) (0.021) (0.050) (0.045) (0.021) (0.024) (0.061) (0.048) (0.019) (0.022) (0.048) (0.046) ln( Net assets ) 0.728 ** 0.741 ** 0.711 ** 0.709 ** 0.701 ** 0.704 ** 0.716 ** 0.683 ** 0.727 ** 0.741 ** 0.711 ** 0.709 ** 0.701 ** 0.703 ** 0.716 ** 0.683 ** (0.007) (0.009) (0.015) (0.012) (0.006) (0.008) (0.015) (0.010) (0.006) (0.009) (0.015) (0.012) (0.006) (0.008) (0.015) (0.010) Two-digit industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country target dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Acquisition year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes R 2 0.700 0.719 0.693 0.694 0.667 0.687 0.664 0.655 0.700 0.720 0.693 0.694 0.667 0.688 0.664 0.656 Observations 13,751 7,805 2,453 3,493 16,555 9,497 2,437 4,621 13,751 7,805 2,453 3,493 16,555 9,497 2,437 4,621 Notes : This table reports the results of OLS regressions that examine the relationship between patents and firm value for different periods. We distinguish between and between-industry acquisitions. Robust standard errors are in brackets. *Significant at 5%. ** Significant at 1%.

(12)

S. Belenzon, A. Patacconi / Research Policy 42 (2013) 1496 – 1510 1507

Domestic-vs.cross-borderacquisitions.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Domesticacquisitions Cross-borderacquisitions Domesticacquisitions Cross-borderacquisitions

All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007

ln(1+USPTOPatentStock)_t₋₁ 0.066** 0.067** 0.045 0.022 0.006 0.015 −0.014 −0.041 0.086** 0.089** 0.066 0.034 0.035 0.043 0.033 −0.048

(0.013) (0.015) (0.039) (0.032) (0.033) (0.044) (0.101) (0.068) (0.012) (0.014) (0.037) (0.031) (0.035) (0.049) (0.082) (0.070) ln(1+EPOPatentStock)_t₋₁ 0.093** 0.092** 0.114** 0.131** 0.155** 0.136** 0.114 0.212** 0.080** 0.074** 0.105** 0.127** 0.134** 0.114* 0.073 0.217** (0.015) (0.017) (0.045) (0.037) (0.036) (0.047) (0.119) (0.076) (0.015) (0.017) (0.042) (0.037) (0.038) (0.051) (0.105) (0.074) ln(Netassets) 0.718** 0.726** 0.709** 0.703** 0.655** 0.659** 0.660** 0.637** 0.717** 0.726** 0.709** 0.703** 0.655** 0.659** 0.659** 0.637** (0.005) (0.006) (0.011) (0.008) (0.014) (0.022) (0.030) (0.025) (0.005) (0.006) (0.011) (0.008) (0.014) (0.022) (0.030) (0.025)

Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

R2 0.691 0.710 0.677 0.680 0.607 0.619 0.672 0.615 0.691 0.710 0.677 0.680 0.607 0.619 0.672 0.615 Observations 26,559 15,536 4,092 6,931 3,747 1,766 798 1,183 26,559 15,536 4,092 6,931 3,747 1,766 798 1,183 Notes:ThistablereportstheresultsofOLSregressionsthatexaminetheeffectofpatentsonﬁrmvaluefordomesticvs.cross-borderacquisitions.Robuststandarderrorsareinbrackets.

**_Signiﬁcant_at_1%.

Table9

American-vs.Europeantargets.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Americantarget Europeantarget Americantarget Europeantarget

All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007

ln(1+USPTOPatentStock)_t₋₁ 0.041** 0.045** 0.007 0.031 0.056** 0.069** 0.070 −0.102 0.061** 0.071** 0.023 0.036 0.083** 0.095** 0.118* −0.123*

(0.016) (0.018) (0.054) (0.047) (0.021) (0.024) (0.064) (0.058) (0.015) (0.017) (0.045) (0.042) (0.021) (0.024) (0.059) (0.062) ln(1+EPOPatentStock)_t₋₁ 0.105** 0.105** 0.145** 0.095* 0.101** 0.096** 0.064 0.216** 0.090** 0.084** 0.136** 0.096* 0.085** 0.077** 0.039 0.227** (0.018) (0.020) (0.064) (0.055) (0.025) (0.029) (0.068) (0.059) (0.018) (0.020) (0.058) (0.054) (0.026) (0.029) (0.065) (0.061) ln(Netassets) 0.719** 0.744** 0.648** 0.684** 0.707** 0.699** 0.741** 0.700** 0.718** 0.742** 0.648** 0.684** 0.707** 0.699** 0.741** 0.700** (0.007) (0.008) (0.020) (0.017) (0.006) (0.009) (0.013) (0.010) (0.007) (0.008) (0.020) (0.017) (0.006) (0.009) (0.013) (0.010)

Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

R2 0.675 0.713 0.563 0.652 0.694 0.697 0.739 0.688 0.676 0.713 0.563 0.652 0.694 0.697 0.740 0.688 Observations 12,001 8,170 1,866 1,965 14,655 8,067 2,476 4,112 12,001 8,170 1,866 1,965 14,655 8,067 2,476 4,112 Notes:ThistablereportstheresultsofOLSregressionsthatexaminetheeffectofpatentsonﬁrmvalue.WesplitthesampledependingonwhethertargetsareAmericanorEuropeanﬁrms.Robuststandarderrorsareinbrackets.